-
如图8所示,实验平台包含一台工业相机(分辨率为1280×1024,帧率为75 fps),一台数字投影仪(分辨率为1280×720 pixel,帧率为60 fps)及一台六轴机械臂(UR5)组成。以带有通孔的铝制环形工件作为被测对象,采用7位互补格雷码,其中,第6位格雷编码条纹宽度为20 pixel。四步相移法采用光栅条纹周期节距为20 pixel。相机与工件内壁间距离约为170 mm,有效视场宽度约为150 mm,实验过程如下:
(1) 将标定板置于旋转平面上,每转动30°,辅助相机2(分辨率为1280×1024,帧率为75 fps)采集1张图像,共采集12张标定图像,拟合面Oa-XaYa和转动平面坐标系原点(Xa0,Ya0,Za0),结果如图9所示,根据公式(6),获取相机2坐标系与转动平面坐标系的变换矩阵P2a,表1为相机2内参与旋转平面坐标系标定结果。
Figure 9. Image of turntable calibration fitting. (a) Checkerboard corners fitting the plane; (b) Checkerboard corners fitting the ellipse
Intrinsic parameters
of camera 1Values Parameters
of camera 1, 2Values fu1 1693.1212 R12 0.8794 0.2368 −0.4131
0.0790 0.7830 0.6170
0.4695 −0.5752 0.6700fv1 1684.9189 u01 617.9206 v01 572.1863 T12 203.2638 −154.7710 −156.0094 k11 0.0165 Table 1. Calibration results of auxiliary camera 2 and turntable
(2) 将相机1、投影仪固定于旋转平面上,对相机1、2进行标定,获取两相机坐标系间的变换矩阵P12。表2为相机1内参和相机1、2坐标系的标定结果。
Intrinsic parameters of camera 2 Values Calibration parameters of turntable Values fu2 1737.9983 R2a −0.0023 −0.0312 0.9951
0.9973 −0.0741 0.0102
0.0741 0.9968 0.0313fv2 1727.9421 u02 639.7128 v02 477.7666 T2a −338.5816 6.4040
−146.7043k12 −0.0179 Table 2. Coordinate transformation relationship of camera 1 internal reference and camera 1, 2
(3) 采用1.3节灰白棋盘格多项式拟合标定方法,采集标定板25个不同位置的光栅及无光栅图像。根据公式(4)获取多项式系数,其中拟合次数N=3。
为了评估文中点云拼接方法的精度,文中对图8中的铝制工件进行了重建:
(1) 将该工件固定到测量位置,投影仪投射条纹编码图案到工件内壁,同时相机采集变形的条纹图像。
(2) 对采集到的条纹图像进行解相处理,获取每个像素的绝对相位值,得出该工件内壁的点云数据。
(3) 通过公式(14)对点云拼接,重建工件内壁三维模型重建结果及细节如图10(a)、(b)所示,相邻角度获取的点云重叠处过渡光滑。取工件内壁圆周4个特征点1、2、3、4的两两间距来对实际数据与测量数据进行比较。表3为4个特征点间连线的实际值与测量值之间的对比。测量值与实际值的平均误差为0.048 mm,标准偏差为0.056 mm,满足拼接精度要求。
Line of measurement Actual value Measured value Error Line12 244.0113 243.9583 0.053 Line13 345.0742 345.0313 0.043 Line14 243.9514 244.0154 −0.064 Line23 244.0342 243.9902 0.044 Lline24 344.9786 345.0336 −0.055 Line34 243.9465 243.9775 −0.031 Average error 0.048 Standard deviation 0.056 Table 3. Comparison between actual value and measured value
Research on rotating splicing of point cloud in workpiece wall based on surface structured light
doi: 10.3788/IRLA20210952
- Received Date: 2021-12-10
- Rev Recd Date: 2022-03-25
- Publish Date: 2022-09-28
-
Key words:
- optical measurement /
- workpiece inner surface /
- point cloud splicing /
- rotating reconstruction
Abstract: In order to solve the problem of limited internal space and difficult measurement of some workpiece, a point cloud rotating splicing method based on surface structured light was proposed. The reconstruction method of single field of surface structured light was introduced in this paper. The absolute phase value was obtained by combining four-step phase shift and complementary Gray code, and the camera and projector were calibrated by polynomial fitting method. The point cloud registration was studied based on the rotation plane of the wrist joint at the end of the manipulator. A calibration method based on the auxiliary camera was proposed, and the transformation relationship between the camera imaging coordinate system and the rotation plane coordinate system was given. The experimental results show that the method is suitable for measuring the inner wall of workpiece, and the average error of splicing is less than 0.05 mm, which meets the requirements of practical application.